Using big data for more dependability

Abstract

There are many large infrastructures that instrument everything from network performance metrics to user activities. However, the collected data are generally used for long-term planning instead of improving reliability and user experience in real time. In this paper, we present our vision of how such collections of data can be used in real time to enhance the dependability of cellular network services. We first discuss mitigation mechanisms that can be used to improve reliability, but incur a high cost which prohibit them to be used except in certain conditions. We present two case studies where analyses of real cellular network traffic data show that we can identify these conditions.

Machine Learning Based Session Drop Prediction in LTE Networks and Its SON Aspects

“…The features of the ML model in these studies are aggregated values of certain radio events and reports in a fixed time window preceding the drop. While the settings greatly differ in these studies, the accuracy of our results is much better than in [3] and comparable to [2]. In both papers, prediction is only made where the session is dropped in the next second.…”

“…For online sessions, several drop avoidance or mitigation mechanisms are proposed in [2], [3], e.g. coordination between voice and data transmissions, switching to other RAT, content pre-fetching, automatic re-connection.…”

“…The authors in [2], [3] use traditional ML models, AdaBoost and Support Vector Machine (SVM), to predict call drops in 3G network and use the prediction result to either avoid them or mitigate their effects. The features of the ML model in these studies are aggregated values of certain radio events and reports in a fixed time window preceding the drop.…”

Mobile Video Delivery: Challenges and Opportunities

“…Our previous work 5 shows another example of how the collaboration between mobile devices and cellular infrastructure can help improve the performance of video streaming. Cellular networks possess rich information about mobile devices' location, mobility patterns, and possible geographic regions where network link qualities might be degraded.…”

Video through a crystal ball

“…Location-based network quality prediction for wireless-LANs has been suggested in [10]. The work in [13] uses real-time cellular data at the base station to predict the quality of the network in the future. Scheduling at the base station based on mobility [9] and shaping video traffic at the the network layer [3,8] are also approaches that can be indirectly used to predict the future network bandwidth.…”